Monday, 18 November 2013

The map is not the territory

Another version of this saying is “Treat the patient, not the X-rays”, but try as we might, we still end up treating the X-rays, even when the evidence is to the contrary.

An X-ray is a shadow of calcium collections (bones) and metal (implants) in our bodies, at one point in time. X-rays don’t tell us what the body part looks like to the naked eye, they don’t tell us how well those parts function, and most of all they don’t tell us if they are causing pain. Yet it is often the images that we treat, rather than the patient.

Wrist fractures, for example
Fractures at the lower end of the radius, often called “Colles” fractures, are the most common fractures seen by doctors. Most of these fractures occur in the elderly, and in women. Previously, such fractures were treated with manual realigning (“closed reduction”) and plaster. Often the alignment was not maintained; the fracture would heal, but with varying deformity at the wrist. Function was usually good after such “malunions” but malunions are now considered failures, necessitating surgical fixation from the start in order to avoid the risk of gradual redisplacement and “malunion” associated with closed reduction and plaster. Various forms of surgical fixation have been used, but what happens these days is that nearly every displaced wrist fracture gets plated, regardless of age.

But making the X-ray look good is not what we are here for.

The best evidence we have for treating displaced wrist fractures in the elderly is from a 2011 randomised trial. In this trial, all fractures were initially treated with a closed reduction and plaster, and then reviewed one week later. Those with fractures that had displaced were randomised either to surgery (to realign the fracture and apply a plate) or continued non-operative treatment (staying in the plaster with the fracture displaced). The results showed that pain and function were similar in both groups. The operated group had much better looking X-rays, but they also had more complications.

There is other evidence that the radiographic appearance of wrist fractures does not correlate with pain or function in older patients, and that surgery does not provide an overall functional advantage in this group (here).

I will grant that plating offers two advantages: earlier movement (less time in plaster) and better grip strength, but plating does not provide an overall advantage in pain or overall function, and all the studies above, and others, have shown consistent complication rates of around 20% from plating.

I recently treated such a patient without surgery, and the fracture healed with noticeable deformity but with good function. The patient was referred to a hand specialist; he and the local doctor were very concerned about the X-rays. In the end, they decided against rebreaking and plating the bone, largely (but not wholly) because the patient had no pain and good function, and was probably wondering what all the fuss was about.

How is this explained?
Radiographs are a surrogate outcome. I have blogged about the problem with surrogate outcomes before, and this is really just another version of the Normalisation Heuristic – our desire to treat what we can measure by correcting it to what we deem “normal”. The link between the X-ray appearance and what the patient’s arm looks like, how well it works and how it feels is weak, at best.

Why do we still base our decisions on images?
  • It is easier
  • It fits the medical model that we can predict the clinical course by investigations
  • It makes sense (if you don’t think about it too hard)
  • Fear of having a bad result (or being sued). It always looks better if the doctor did something (at least he tried), rather than treating the condition non-operatively (“he didn’t do anything!”). (see my previous post on doctors’ bias towards doing something rather than nothing)
  • It is very difficult for doctors to accept fracture displacement (or indeed, any abnormality) on an X-ray when they know that it can be corrected (regardless of whether it needs correction). Again, see previous post on this topic
  • It is also hard for the patient to accept a bad looking X-ray
  • It is what the doctor has been taught
  • Everyone else is doing the same thing

Other examples
Treating degenerative changes in the shoulder (rotator cuff) and spine are more examples of falsely attributing the (very common) symptoms of shoulder, neck and back pain to the (very common) changes seen on images. Treating “torn” (usually degenerative) menisci in arthritic knees is another. Many other fractures do very well when left alone (for example, 5th metacarpal fractures in the hand), There are many other examples, but suffice it to say, in many institutions these days patients are automatically booked for surgery when the image of a displaced fracture is seen.

The bottom line

The risks of plating displaced wrist (distal radius) fractures in the elderly are underestimated and the benefits over alternative treatments are overestimated. It is one example of equating the surrogate with reality; of treating the images and not the patient.


  1. Even in regards to fixation - there is a good recent paper to show the type of fixation is less relevant - - again short term outcomes and xray outcomes are better in the plating group, however return to work and function seem largely unchanged

    1. Thanks. I originally included that article, but as it compares plating to other forms of fixation I left it out. There are other similar papers. I think the articles that compare plating to plaster are more revealing: instead of arguing over which form of surgical fixation should be used, we should be asking whether surgery is necessary for many of these fractures.

  2. I wish someone would apply the same logical process to the "Back to Sleep" programme wrt infant plagiocephaly.

    1. Thanks. I just briefly looked this up. Some are blaming the Back to Sleep program for cranial deformation. Like so much medicine: high on speculation and low on good science.

  3. This comment has been removed by the author.

  4. Imaging displaces touch as an application of defensible medicine. The x-ray can be preserved and presented as evidence. Perception based on observation and touch is more difficult to certify. Why is it so difficult to leave good enough alone when we can break it again to make it look better?

    1. Thanks, and I think you have touched on the crux of the problem: we are more likely to rely on objective, reproducible measures, which is not a bad thing, except when they are inferior surrogates of the real thing.

  5. We need a list of clinically meaningful treatments. Then when confronted by a patient we can show if you have "x" there is actually something we can do that will help and not actually make you worse. If you have a condition that will not benefit from medical intervention then it isn't offered and it isn't done.....unless you want to pay for it yourself.